compare: show aggregated diff of what will be merged to other repo, using merge ancestor

pull_request.get_compare_data will also now return the ancestor that would be
used for actual merging. Showing a diff from that ancestor instead of the first
'new' changeset will give a more realistic diff that doesn't include merges.

rhodecode/controllers/compare.py

@@ -126,9 +126,8 b' class CompareController(BaseRepoControll'
             org_ref = ('rev', rev_start)
             other_ref = ('rev', rev_end)
 
-        c.cs_ranges = PullRequestModel().get_compare_data(
-                                    org_repo, org_ref, other_repo, other_ref,
-                                    )
+        c.cs_ranges, ancestor = PullRequestModel().get_compare_data(
+            org_repo, org_ref, other_repo, other_ref)
 
         c.statuses = c.rhodecode_db_repo.statuses([x.raw_id for x in
                                                    c.cs_ranges])
@@ -141,15 +140,14 b' class CompareController(BaseRepoControll'
         c.org_ref = org_ref[1]
         c.other_ref = other_ref[1]
 
-        if c.cs_ranges and c.org_repo != c.other_repo:
-            # case we want a simple diff without incoming changesets, just
-            # for review purposes. Make the diff on the forked repo, with
+        if ancestor  and c.org_repo != c.other_repo:
+            # case we want a simple diff without incoming changesets,
+            # previewing what will be merged.
+            # Make the diff on the forked repo, with
             # revision that is common ancestor
             _org_ref = org_ref
-            org_ref = ('rev', getattr(c.cs_ranges[0].parents[0]
-                                      if c.cs_ranges[0].parents
-                                      else EmptyChangeset(), 'raw_id'))
-            log.debug('Changed org_ref from %s to %s' % (_org_ref, org_ref))
+            log.debug('Using ancestor %s as org_ref instead of %s', ancestor, _org_ref)
+            org_ref = ('rev', ancestor)
             org_repo = other_repo
 
         diff_limit = self.cut_off_limit if not fulldiff else None

rhodecode/lib/diffs.py

@@ -713,4 +713,4 b' def differ(org_repo, org_ref, other_repo'
             ignore_whitespace=ignore_whitespace, context=context)
         return _diff
 
-    return ''
+    return '' # FIXME: when is it ever relevant to return nothing?

rhodecode/model/pull_request.py

@@ -160,8 +160,8 b' class PullRequestModel(BaseModel):'
 
     def _get_changesets(self, alias, org_repo, org_ref, other_repo, other_ref):
         """
-        Returns a list of changesets that are incoming from org_repo@org_ref
-        to other_repo@other_ref
+        Returns a list of changesets that can be merged from org_repo@org_ref
+        to other_repo@other_ref ... and the ancestor that would be used for merge
 
         :param org_repo:
         :param org_ref:
@@ -170,7 +170,7 b' class PullRequestModel(BaseModel):'
         :param tmp:
         """
 
-        changesets = []
+        ancestor = None
 
         if alias == 'hg':
             # lookup up the exact node id
@@ -202,9 +202,14 b' class PullRequestModel(BaseModel):'
 
             revs = ["ancestors(id('%s')) and not ancestors(id('%s'))" %
                     (other_rev, org_rev)]
-            out = scmutil.revrange(hgrepo, revs)
-            for cs in (out):
-                changesets.append(other_repo.get_changeset(cs))
+            changesets = [other_repo.get_changeset(cs)
+                          for cs in scmutil.revrange(hgrepo, revs)]
+
+            if org_repo != other_repo:
+                ancestors = scmutil.revrange(hgrepo,
+                     ["ancestor(id('%s'), id('%s'))" % (org_rev, other_rev)])
+                if len(ancestors) == 1:
+                    ancestor = hgrepo[ancestors[0]].hex()
 
         elif alias == 'git':
             assert org_repo == other_repo, (org_repo, other_repo) # no git support for different repos
@@ -212,11 +217,10 b' class PullRequestModel(BaseModel):'
                 'log --reverse --pretty="format: %%H" -s -p %s..%s' % (org_ref[1],
                                                                        other_ref[1])
             )
-            ids = re.findall(r'[0-9a-fA-F]{40}', so)
-            for cs in (ids):
-                changesets.append(org_repo.get_changeset(cs))
+            changesets = [org_repo.get_changeset(cs)
+                          for cs in re.findall(r'[0-9a-fA-F]{40}', so)]
 
-        return changesets
+        return changesets, ancestor
 
     def get_compare_data(self, org_repo, org_ref, other_repo, other_ref):
         """
@@ -242,7 +246,7 b' class PullRequestModel(BaseModel):'
         other_repo_scm = other_repo.scm_instance
 
         alias = org_repo.scm_instance.alias
-        cs_ranges = self._get_changesets(alias,
-                                         org_repo_scm, org_ref,
-                                         other_repo_scm, other_ref)
-        return cs_ranges
+        cs_ranges, ancestor = self._get_changesets(alias,
+                                                   org_repo_scm, org_ref,
+                                                   other_repo_scm, other_ref)
+        return cs_ranges, ancestor